Document OCR
Document OCR returns the text of documents without adding them to the Library. It runs on Rivul AI's own server and needs no AI key.
Extract text
Step 1: Open Document OCR in the sidebar, under Convert (rivul.ai/app/ocr).
Step 2: Drop, upload, or paste up to 10 files: PDFs (with a text layer or scanned), PNG, JPEG, TIFF, or WebP, up to 25 MB each. Files are read one at a time.
Step 3: Select a file to see its text. The Text tab gives plain text; the Markdown tab keeps headings and tables. A document longer than 50 pages returns its first 50 pages and says so.
Step 4: Copy the result, download it as .txt or .md, or, for a PDF, click Add to Library, which uses the normal library upload and its limits.
Privacy and accuracy
Each file is read in memory, deleted from temporary storage after reading, and not stored or logged. No AI provider or language model is involved.
OCR can misread digits, symbols, and table cells, so check numbers and quotations against the original before citing them.
Scanned PDFs in the Library
A scanned PDF you add to the Library is read with OCR automatically, so you only need Document OCR when you want text without storing the file. See Library for extraction status and Re-extract.
Something here wrong or missing? Use the contact form.